Method and electronic device for controlling operation of a self driving car

Abstract

A method and electronic device for controlling operation of a Self Driving Car (SDC) are disclosed. The method includes, at a given timestamps during operation of the SDC when located at a current location: generating a candidate location of the SDC using a localization algorithm, generating a parameter using a Machine Learning Algorithm (MLA) indicative of whether the localization algorithm is likely to generate divergent candidate locations for the SDC in a candidate portion of the map representation under a variety of conditions, determining that the localization algorithm to be an unreliable localization source in the candidate portion of the map representation using the parameter, determining a multi-sourced location of the SDC using data acquired from a reduced set of localization sources, the reduced set of sources excludes the localization algorithm, and controlling operation of the SDC using the multi-sourced location as the current location of the SDC.

1 . A method of controlling operation of a Self-Driving Car (SDC), the SDC being communicatively coupled to a Light Detection and Ranging (LIDAR) system and an electronic device configured to acquire data from a plurality of localization sources for locating the SDC on a map representation of a geographical region, the method comprising:
 at a given timestamps during operation of the SDC when the SDC is located at a current location in the geographical region:
 generating, using a localization algorithm, a candidate location of the SDC using a point cloud captured by the LIDAR system, the map representation, and an initial approximation of the current location of the SDC on the map representation; 
 generating, using a Machine Learning Algorithm (MLA), a parameter indicative of whether the localization algorithm is likely to generate divergent candidate locations for the SDC in a candidate portion of the map representation under a variety of conditions, the candidate portion including the current location; 
 determining, using the parameter, that the localization algorithm to be an unreliable localization source in the candidate portion of the map representation; and 
 determining a multi-sourced location of the SDC using data acquired from a reduced set of localization sources, the reduced set of localization sources excluding the localization algorithm; and 
 controlling operation of the SDC using the multi-sourced location as the current location of the SDC. 
   
     
     
         2 . The method of  claim 1 , wherein the candidate location is a candidate position of the SDC. 
     
     
         3 . The method of  claim 1 , wherein the candidate location is a candidate position and a candidate orientation of the SDC. 
     
     
         4 . The method of  claim 1 , wherein the variety of conditions include at least one of rain, snow, and dirt occluding the LIDAR system. 
     
     
         5 . The method of  claim 1 , wherein the map representation is a High Definition (HD) map. 
     
     
         6 . The method of  claim 1 , wherein the set of localization sources comprises an odometry based localization source, a LIDAR based localization source, image-based localization source, and Global Navigation Satellite System (GNSS) source. 
     
     
         7 . An electronic device for controlling operation of a Self-Driving Car (SDC), the SDC being communicatively coupled to a Light Detection and Ranging (LIDAR) system and the electronic device configured to acquire data from a plurality of localization sources for locating the SDC on a map representation of a geographical region, the electronic device being configured to:
 at a given timestamps during operation of the SDC when the SDC is located at a current location in the geographical region:
 generate, using a localization algorithm, a candidate location of the SDC using a point cloud captured by the LIDAR system, the map representation, and an initial approximation of the current location of the SDC on the map representation; 
 generate, using a Machine Learning Algorithm (MLA), a parameter indicative of whether the localization algorithm is likely to generate divergent candidate locations for the SDC in a candidate portion of the map representation under a variety of conditions, the candidate portion including the current location; 
 determine, using the parameter, that the localization algorithm to be an unreliable localization source in the candidate portion of the map representation; and 
 determine a multi-sourced location of the SDC using data acquired from a reduced set of localization sources, the reduced set of localization sources excluding the localization algorithm; and 
 control operation of the SDC using the multi-sourced location as the current location of the SDC. 
   
     
     
         8 . The electronic device of  claim 7 , wherein the candidate location is a candidate position of the SDC. 
     
     
         9 . The electronic device of  claim 7 , wherein the candidate location is a candidate position and a candidate orientation of the SDC. 
     
     
         10 . The electronic device of  claim 7 , wherein the variety of conditions include at least one of rain, snow, and dirt occluding the LIDAR system. 
     
     
         11 . The electronic device of  claim 7 , wherein the map representation is a High Definition (HD) map. 
     
     
         12 . The electronic device of  claim 7 , wherein the set of localization sources comprises an odometry based localization source, a LIDAR based localization source, image-based localization source, and Global Navigation Satellite System (GNSS) source.